I’ve posted a few days ago, asking how to setup my storage for Proxmox on my Lenovo M90q, which I since then settled. Or so I thought. The Lenovo has space for two NVME and one SATA SSD.
There seems to a general consensus, that you shouldn’t use consumer SSDs (even NAS SSDs like WD Red) for ZFS, since there will be lots of writes which in turn will wear out the SSD fast.
Some conflicting information is out there with some saying it’s fine and a few GB writes per day is okay and others warning of several TBs writes per day.
I plan on using Proxmox as a hypervisor for homelab use with one or two VMs runnning Docker, Nextcloud, Jellyfin, Arr-Stack, TubeArchivist, PiHole and such. All static data (files, videos, music) will not be stored on ZFS, just the VM images themselves.
I did some research and found a few SSDs with good write endurance (see table below) and settled on two WD Red SN700 2TB in a ZFS Mirror. Those drives have 2500TBW. For file storage, I’ll just use a Samsung 870EVO with 4TB and 2400TBW.
| SSD | TB | TBW | € |
|---|---|---|---|
| 980 PRO | 1TB | 600 | 68 |
| 2TB | 1200 | 128 | |
| SN 700 | 500GB | 1000 | 48 |
| 1TB | 2000 | 70 | |
| 2TB | 2500 | 141 | |
| 870 EVO | 2TB | 1200 | 117 |
| 4TB | 2400 | 216 | |
| SA 500 | 2TB | 1300 | 137 |
| 4TB | 2500 | 325 |
Is that good enough? Would you rather recommend enterprise grade SSDs? And if so, which ones would you recommend, that are m.2 NVME? Or should I just stick with ext4 as a file system, loosing data security and the ability for snapshots?
I’d love to hear your thought’s about this, thanks!
Don’t sweat it.
I remember looking into this as well like a year ago. I also found the same info and started to look into ssds, consumer and enterprise grade and after all that I realised that most of it is just useless fuzzing about. Yes it is an interesting rabbit hole in which I spent a week probably. In the end one simple thing nullifies most of this: you can track writes per day and SSD health. It is not like you need to somehow made a guess when the drives fail. You do not. Keep track of the health and writes per day and you will get a good sense of how your system behaves. Run that for 6 months and you are infinitely wiser when it comes to this stuff.
That rabbit hole is interesting, but also deep and scary. I’m trying to challenge myself by setting up Proxmox, as so far I’ve just used Raspbery Pis as well as OpenMediaVault. So when I saw those stories about drives dying after 6 months, I was a bit concerned;. Especially because I can’t yet verify the truth in those storries, since I’d call myself and advanced novice if I’, being generous.
I’ll track drive usage and wear and see what my system does. Good point, then I can get rid of the guesswork. Thank you a lot!
I’ll agree with the other commenter here.
Also there may not be any difference between the consumer and enterprise drives. The reason the enterprise cost more is the better warranty. But because they have different components.
Monitor the drives, modern drives are pretty good at predicting when they are dying, and replace it necessary.
Yeah, concering TBW there wasn’t a huge difference between cosumer- and enterprise drives that I saw. Something along 2500TBW vs. 3500TBW (unless you go with those unaffordable drives, then yes). I’ll monitor the drives and if I find rapidly increasing wear, I can still switch to another file system. The whole reason I bought the Lenovo is to setup a second machine and experiment, while I still have a running “production” system. Thank you!
ZFS doesn’t eat your SSD endurance. If anything it is the best option since you can enable ZSTD compression for smaller reads/writes and reads will often come from the RAM-based ARC cache instead of your SSDs. ZFS is also practically allergic to rewriting data that already exists in the pool, so once something is written it should never cost a write again - especially if you’re using OpenZFS 2.2 or above which has reflinking.
My guess is you were reading about SLOG devices, which do need heavier endurance as they replicate every write coming into your HDD array (every synchronous write, anyway). SLOG devices are only useful in HDD pools, and even then they’re not a must-have.
IMO just throw in whatever is cheapest or has your desired performance. Modern SSD write endurance is way better than it used to be and even if you somehow use it all up after a decade, the money you save by buying a cheaper one will pay for the replacement.
I would also recommend using ZFS or BTRFS on the data drive, even without redundancy. These filesystems store checksums of all data so you know if anything has bitrot when you scrub it. XFS/Ext4/etc store your data but they have no idea if it’s still good or not.
Thank you so much for this explanation. I am just a beginner, so those horror stories did scare me a bit. I also read, that you can fine tune ZFS to prevent write amplification so I’ll read into that subject a bit more.
I thought ZFS without redundancy did give no benefits, but I most have gotten that wrong. Thanks again!
ZFS without redundancy is not great in the sense that redundancy is ideal in all scenarios, but it’s still a modern filesystem with a lot of good features, just like BTRFS. The main problem will be that it can detect data corruption but not heal it automatically. Transparent compression, snapshotting, data checksums, copy-on-write (power loss resiliency), and reflinking are modern features of both ZFS/BTRFS, and BTRFS additionally offers offline-deduplication, meaning you can deduplicate any data block that exists twice in your pool without incurring the massive resources that ZFS deduplication requires. ZFS is the more mature of the two, and I would use that if you’ve already got ZFS tooling set up on your machine.
Note that the TrueNAS forums spread a lot of FUD about ZFS, but ZFS without redundancy is ok. I would take anything alarmist from there with a grain of salt. BTRFS and ZFS both store 2 copies of all metadata by default, so bitrot will be auto-healed on a filesystem level when it’s read or scrubbed.
Edit: As for write amplification, just use ashift=12 and don’t worry too much about it.
Where can I read more about good ZFS settings for a filesystem on a new RAID6 array? I don’t want to manage disks or volumes with ZFS, I’ll be doing that with mdadm, just want ZFS as filesystem instead of ext4. I assume a ZFS filesystem can grow if the space available expands later?
I barely scratched the surface with ZFS, so I’m not going to touch another file system for a while now. I’m fine with detecting data corruption only, since those files (on the static data storage) can be replaced easily and hold no real value for me. All other data will be either on the redundant pool or is saved to several other media and even one off-site copy.
I already wrote down ashift=12 in my notes for when I set it up.
In general, I found there is a lot of FUD out there when it comes to data security. One I liked a lot was ECC RAM being mandatory for ZFS. Then one of the creators of it basically said: "Nah, it’s not needed more than for any other file system’.
I’m kinda repeating things already said here, but there’s a couple of points I wanted to highlight…
Monitor the SMART health: Enterprize and consumer drives fail, it’s good to know in advance.
Plan for failure: something will go wrong… might be a drive failure, might be you wiping it by accident… just do backups.
Use redundancy; several cheapo rubbish drives in a RAID / ZFS / BTRFS pool are always better than 1 “good” drive on it’s own.
Main point: build something and destroy it to see what happens, before you build your “final” setup - experience is always better than theory.
I built my own NAS and was going with ZFS until I fkd around with it… for me… I then went with BTRFS because of my skills, tools I use, etc… BTRFS just made more sense to me… so I know I can repair it.
And test your backups 🎃
I’m currently playing around in VMs even before I order my hard drives. Just to see, what I can do. Next up is to simulate a root drive failure and how to replace that. I also want to test rolling back from snapshots.
The data that I really do need and can’t replace is redundant anyway: one copy on my PC, one on my external HDD, one on my NAS and one on a system at my sisters place. Thats 4 copies on several media (one cold) and at another place. :)
